Genetic Descent

[Larry Trask]

--On Wednesday, May 23, 2001 11:48 am -0500 "David L. White"
<dlwhite at texas.net> wrote:

[from Thomason and Kaufman]

>         Well, I can, on p. 14:  "... any linguistic feature can be
> transferred from any language to any other language."  No restrictions are
> mentioned, and TK consistently fail to note that not everything that could
> conceivably occur actually does.

Ah, I see.  But this statement seems very different from the position you
imputed to T and K: "Anything goes."  This quote plainly does not say that
anything can happen at all, but only that there are no linguistic features
that cannot be borrowed.  Are you aware of any features that cannot be
borrowed?

[on Thomason's agreement that Anglo-Romani is a variety of English]

>         So even the mixed language crowd now admits that a language which
> has grammar from source A and (non-grammatical) lexicon from source B is a
> form of language A?  That is pretty much what I have been saying.

No; this is not at all what Thomason is saying.  She is saying that
Anglo-Romani is a form of English, and no more.  She is not conceding or
asserting any general point at all.  And it seems quite clear that, in
other cases, such as that of Laha, she concludes that a language with
grammar from source A but lexis from source B is in fact a variety of B.

By the way, I am troubled by this seemingly dismissive expression "the
mixed-language crowd".  Are you suggesting that no mixed languages can
exist at all?  If so, let me ask this: in what respect does Michif fall
short of being a mixed language?  It looks to me like a paradigm case of a
mixed language.  If we encountered, or imagined, a real mixed language,
what features would it have that Michif lacks?

>> [on my example of Takia, which has retained Austronesian morphemes but
>> borrowed grammatical patterns wholesale from the Papuan language Waskia]

>>>         It is Austronesian, and not much of a puzzle, except for TK, who
>>> must try to figure out, without any very clear standard, whether it is
>>> 1) Austronesian, 2) Papuan, or 3) a new "mixed" language.

[LT]

>> First, why is it Austronesian?  By what criterion does the origin of
>> morphemes wholly outweight the origin of morphological patterns?  Isn't
>> this rather arbitrary?

>         Categories can be transferred rather easily.  We do not call Old
> Lithuanian a mixed langage between Baltic and Finnic, or a form of Finnic,
> merely because it created (if I am remembering correctly) an allative.
> All this means is that the language was evidently imposed on a large
> number of Finnic speakers at some point.  (That there is substantial
> Finnic or Uralic sub-stratal influence in Baltic and Slavic is asserted
> by TK, by the way, though I do not recall that they mention this
> partiuclar example.)  Mere borrowing of categories does not affect
> genetic descent.

But the Takia case is not similar to that of Lithuanian.  Takia has not
merely acquired a single new grammatical form by contact: instead, it has
imported the *entirety* of the Waskia morphological system, leaving behind
no original Takia morphological patterns at all.  In a strong sense, it has
imported the entire verbal (and nominal) morphology of Waskia, even though
it has not borrowed any morphemes.  Does this not make a very big
difference?

Consider a similar but hypothetical case.  Suppose English-Navaho
bilinguals were to create a new speech variety, consisting wholly of
English morphemes arranged wholly in Navaho grammatical patterns, with
Navaho grammatical distinctions but not English ones, and the long Navaho
sequences of morphemes but no English sequences.  By your reasoning, the
result would be clearly a form of English and not a form of Navaho.  Is
this reasonable?

>> Second, why is it incumbent upon T and K to classify Takia, or any
>> language, in such a rigorous way?  Why aren't they free to decide "none
>> of the above"?

>         What would "none of the above be"?  The three possibilities given
> above exhaust those that are reasonable.

The missing possibility is this: it makes little sense to ask the question
in the first place.  It may well be that rigid categorization is not
appropriate for (some) mixed languages.  The choice among asserting that
Takia "is" Austronesian, or "is" Papuan, or "is" neither may simply be
inadequate to capture linguistic reality.

We've just had the census here in Britain.  The census form tries
desperately to classify every individual into one ethnic group or another.
But not everybody fits very well.  What ethnic group does Tiger Woods
belong to?  In fact, I understand, Mr. Woods, who clearly has a sense of
humor, has invented a new ethnic group for himself -- an ethnic group of
which he is possibly the only member.

Why should languages be simpler than people?

> And since when is the equivalent
> of a mental coin toss to be preferred to a clear standard?

No one is proposing a coin toss.  Anyway, what is the value of having "a
clear standard" if that standard forces us into conclusions which fly in
the face of the facts?  You have picked verbal morphology as your standard,
and now you have further selected the etymology of morphemes in preference
to the origin of morphological patterns.  All those seems wholly arbitrary
to me, and in no way to be preferred to any other arbitrary criterion.

> Since
> borrowing of categories, without borrowing of specific morphmes to
> express these, is not relevant, the case in question is practically
> indentical to the case of Anglo-Romani.

Who says that the borrowing of categories is not relevant?  It appears that
you have just declared this by fiat.  Languages borrow words and morphemes
at least as often as they borrow grammatical categories.  So why is the
etymology of morphemes sacrosanct, but not the etymology of grammatical
categories and patterns?

But now consider another case discussed by Thomason: Kormakiti Arabic.
This variety is the mother tongue of the people in one village in Cyprus.
It is descended from the Arabic brought to the island in the twelfth
century.  It consists of a complex mixture of Arabic and Cypriot Greek.
The lexicon, based on a sample, is about 62% Arabic and 38% Greek, with the
Greek component including many items of basic vocabulary.  Arabic words are
built from Arabic phonemes and take Arabic morphology and phrasal syntax.
Greek words are built from Greek phonemes and take Greek morphology and
phrasal syntax.  Phrases are combined into sentences by rules which are a
mixture of Arabic and Greek.

So, in this language, Arabic verbs take Arabic verbal morphology, while
Greek verbs take Greek verbal morphology.  What does your "clear standard"
say about this case?

[LT]

>> Anyway, I now want to talk about another language, discussed by Thomason
>> in her new book.  That language is Laha, spoken in the Moluccas, where
>> the locally dominant language is a variety of Malay, which is distantly,
>> but not closely, related to Laha.  The principal investigator here is
>> James T. Collins:

>>   J. T. Collins. 1980. 'Laha, a language of the central Moluccas'.
>>   Indonesia Circle 23: 3-19.

>> The lexicon of Laha is largely native, though partly borrowed from Malay.
>> But the grammar is almost 100% Malay, with only a few fragments of native
>> Laha grammar surviving.  So, my question for David White is this: is
>> Laha a variety of Malay, or not?  Since the grammar is almost entirely
>> Malay, it would seem that he must answer "Yes; it's Malay."

>> But now consider the following.  The Laha are reportedly regarded by
>> themselves, and by others, as a distinct ethnic group with a distinct
>> language.  All Laha-speakers are also fluent in Malay, but other speakers
>> of Malay do not speak Laha.

>         The Romani are regarded by themselves, and by others, as a
>         distinct ethnic group with (where it survives) a distinct
> language.  All Anglo-Romani speakers are fluent in English, but other
> speakers of English do not speak Anglo-Romani.  Why come to a different
> conclusion in the two cases, unless coin tossing is the latest fashion?

What is this thing with coin-tossing?  No one is suggesting any such thing.
What T and K are suggesting is that every case of contact is different, and
that each case must be carefully examined before any conclusions can be
drawn: we cannot simply decide in advance, by fiat, what is possible and
what is not.  Is that unreasonable?

>> The conclusion of Collins, and of Thomason, is
>> that Laha has developed as follows: it started out as a language entirely
>> distinct from Malay, but, under enormous pressure from Malay, it has
>> gradually, in piecemeal fashion, absorbed more and more Malay grammar,
>> until today the grammar is almost entirely Malay, and only a few bits of
>> Laha grammar survive -- for the moment, since it is possible that these
>> few fragments will also give way to Malay grammar in the future.

>> Therefore, using David White's criterion, Laha has changed from being a
>> language entirely distinct from Malay to a mere variety of Malay.

>         If the fantasy scenario envisaged above is to be taken as a fact.

Who says this is "a fantasy scenario"?  The investigator has looked
carefully at Laha and concluded that it arose by massive replacement of the
original Laha grammar by Malay grammar.  Why does this unremarkable
conclusion bother you so much?

>> Is this
>> reasonable?  Earlier, David rejected a similar scenario for another
>> language as impossible in principle.

>> So, the possible conclusions:

>> (1) The speech variety called 'Laha' has indeed changed from being one
>> language to being another.

>> (2) Laha has always been, and remains, a language distinct from Malay,
>> even though it has imported almost the entirety of Malay grammar.

>         It is a language distinct from Malay but descended from Malay.
> Sudden re-lexification, as probably happened in both Anglo-Romani and
> Mednyj Aleut, can create this.

How do you know there was "sudden relexification"?  Isn't this just another
"fantasy scenario"?

> Such developments are unusual and should
> be recognized as distinct from the more normal sort of genetic descent,
> where mutual intelligibility across generations always exists, but, as
> Thomason in effect admits in her interpretation of Anglo-Romani, do not
> constitute non-genetic descent.

Now you are proposing that a language can arise so suddenly that there is
no mutual comprehensibility across generations?  This strikes you as more
plausible than gradual regrammaticalization?  Whew.

>> (3) Collins and Thomason are completely wrong in their interpretation,
>> and Laha must have had some other origin.

>         Probably.  If piecemeal borrowing of parts verbal morhpology,
>         which woyld seem to be possible under Thomason's scenario, is
> possible, we should (hopefully) be able to find some examples of the
> process caught in the act, resulting in mixed verbal morphology.  Unless
> Laha is one (and so far I have seen no evidence to show that it is), I
> can still assert that there are not any.

All right -- Kormakiti Arabic, then.

[snip]

>         The scenario envisaged by Collins and Thomason is just that:  a
> scenario envisaged.  It is not a fact.  I would need to know what the "few
> bits of Laha grammar" are.  If they involve finite verbal morphology, then
> what I have been saying is wrong.  (Though pointing out that TK's examples
> did not in fact show this would still hardly be an outrage.)  If they
> involve nominal morphology, derivational or non-finite verbal morphology,
> or (worse yet) mere categories, then what I have been saying is not wrong.

I regret that I can't answer this, since Thomason doesn't provide that much
detail, and since our library doesn't take the journal in which Collins's
article appears.  However, I'm intrigued about this case now, so I'll see
if I can find out.  Unfortunately, I am now buried in the annual ordeal of
exam marking, so I won't be doing anything quickly.

> Malay, by the way, is by typology if not history a semi-creole, so I would
> also like to know what the parts of "Malay grammar" that have supposedly
> been imported supposedy are.

To be honest, I can't see why this is relevant.  The point is, in Collins's
interpretation, that Laha grammar was originally different from Malay
grammar but no longer is, except in a few details.

> But, operating in the semi-dark (Thomason is
> routinely vague about the facts of her example languages), I see no reason
> to think that the situation is not comparable to Anglo-Romani or Mednyj
> Aleut, which are surely not mutually comprehensible with English or
> Russian respectively (or for that matter Romani or Aleut), but which
> nonetheless can be identified (even by Thomason, in the first case) as
> forms of "abruptly re-lexified" English or Russian.

I am not so sure, but I'll see what I can find out, when the last pile of
scripts disappears from my desk.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk

Tel: (01273)-678693 (from UK); +44-1273-678693 (from abroad)
Fax: (01273)-671320 (from UK); +44-1273-671320 (from abroad)